I've just found this thread, but I'll throw out a few observations to stir the pot...
There is nothing particularly revolutionary about Java. As a language it is little more than C++ made safe. This is a good thing, but not in itself world-shaking. Even the cross platform portability is nothing particularly new... anyone like to compare the number of platforms supporting Java with the number supporting COBOL? This is not to say that Java doesn't have some compelling aspects, but a great deal of the Java dream depends on the success of the write once, run anywhere concept. I think this concept is significantly at risk because of the current Microsoft - Sun wars. It may be exciting to dream about all the optional alternatives to Wintel platforms which Java could help support, but the fact of the matter today is that a developer wanting to sell a desktop application is almost certain to write it first for Wintel. If MSFT succeeds in getting that developer to use the MS class libraries instead of the Sun ones, how many do you think will bother developing the second version? I'm not making this a prediction, but it is at least a big cloud.
I also have to wonder about the whole NC fad. All of the NC offerings I have seen so far are barely cheaper than an equivalently equipped PC, if that. To be sure, NC pricing will drop as production volumes increase, but so will PC prices, especially as power continues to go through the roof and many users will find that last year's model provides more than enough for their needs. I see the NC making inroads in the same places that X terminals had a limited bloom, i.e., controlled corporate environments where cost of overall administration was a big issue and it was desired to have only the provided applications available. But, there are lots of people who will continue to be willing to pay a few dollars more to get a general purpose machine that will run just about any software they want. We may get to a spot soon where there are acceptable solutions for NCs for certain general computing requirements, but it will be a *long* time before there is anything like the same range of software as there is for Wintel.
I think we need to remember too that, despite the level of turnover in PCs these days, these are not disposable items that get used up and replaced every few years. I still have customers running on 286s! Even a dramatic upsurge in sales of NCs would still take many years before getting so far as to reach parity with the installed base of PCs, especially since many new customers will continue to buy PCs in those circumstances in which they are the preferred solution. E.g., how exactly is it "better" to be connected to the Internet 40 hours a week just to be able to work when a standalone PC with occassional access will do? Even when and if cable modems become universally available, they are no faster at best than the current slow ethernet LAN technology, the very thing which is cited as a performance bottleneck in many environments.
My own feeling is that what we are most likely to see in the short run is a period of increasing diversification in the choice of alternatives for the desktop, following a period in which we have been seeing a decrease. This will stimulate demand for platform transportable solutions, in which Java will be a player, but only a part of the overall picture. As for the long run, the one most sure thing about this biz is that any long run predictions are almost certain to be wrong, not just because of the complexity of the forces, but because the technology itself will change in sufficiently fundemental ways that the impact cannot be foreseen. |
